Increased duration of mechanical ventilation is associated with decreased diaphragmatic force: a prospective observational study

Patients were eligible if they were admitted between March 2007 and October 2008 to the medical intensive care unit, which is a 17-bed ICU of a tertiary center with approximately 750 admissions yearly. Inclusion criteria consisted of intubation and mechanical ventilation, hemodynamic stability, and stable respiratory status with a positive end-expiratory pressure (PEEP) ≤7 cm H ₂O. Contraindications were those related to magnetic stimulation (pacemaker or implantable cardioverter-defibrillator, prosthetic valve, cervical implants, cervical trauma), contraindication for gastric/esophageal balloon placement (coagulation disorders, low platelet count, gastric or esophageal pathology), factors possibly interfering with correct pressure measurements (multiple-functioning chest drains, severe abdominal infections, recent major abdominal surgery, agitation, bronchospasm), age younger than 18 years, pregnancy, and having received neuromuscular blocking agents in the past 24 h.

Mechanical ventilators used were Dräger, EvitaXL, and Dräger, Evita4. During mechanical ventilation, the need for analgesics and sedatives was daily evaluated and titrated by the treating physician to the minimum needed, aiming at a sedation agitation score [ 21] of 3 to 4. Informed consent was obtained from the patients or their relatives. The study was approved by the local ethics committee.

We measured twitch transdiaphragmatic pressure (TwPdi) by using bilateral anterior magnetic phrenic nerve stimulation (BAMPS), as described previously [ 19]. In brief, the phrenic nerves were stimulated bilaterally from the anterior approach, at the posterior border of the sternocleidomastoid muscle, at the level of the cricoids, by using two figure-of-eight 45-mm magnetic coils (Magstim, Dyfed, Wales) and a bistim (Magstim, Dyfed, Wales). A custom-built, two-way occlusion valve was used to create isometric conditions during stimulation. Esophageal and abdominal pressure changes were measured by using balloon catheters (UK Medical, Sheffield, UK) inserted through the nose after local anesthesia. The gastric balloons were filled with 2 ml of air, and the esophageal balloons contained 0.5 ml of air. To verify correct positioning of the abdominal catheter, abdominal compression was applied. The position of the esophageal catheter was confirmed to be correct if the end-expiratory pressure was near the PEEP level applied, and if inspiration against the closed airway resulted in similar pressure changes on the esophageal and tracheal tracings. Tracheal, abdominal (TwPabd), and esophageal pressures (TwPes) were measured by using Validyne MP45 transducers, 250 cm H ₂O, connected to a custom-built carrier amplifier. Tracheal pressure was measure at the proximal end of the endotracheal tube. Biopac MP150 (Cerom, Paris, France) was used as the data-acquisition system. Patients were left breathing quietly for 20 min after balloon placement. After determining the optimal position of the coils, at least three stimulations were performed at 100% of maximal output. To evaluate supramaximality, patients were also stimulated at 70%, 90%, and, if necessary, at 95%. All measurements were performed with the head end in 30-degree upward position. Between two stimulations, at least a 30-sec pause was obtained to avoid superposition. To evaluate repeatability of the measurement, if possible, patients were measured on two occasions as close together as technically feasible and according to the patients' clinical status, but with a minimal time interval of 24 h.

Individual abdominal and esophageal pressure signals were accepted for analysis if they were timed at end expiration with no major cardiac artefact, if stable and physiologically acceptable end-expiratory pressure was present, and if active abdominal muscle contraction or esophageal contraction during the stimulation was absent. TwPes and Twabd were defined as the maximal excursion of the esophageal and abdominal tracing, respectively, on stimulation from the value immediately before stimulation (Figure 1). TwPdi was calculated as the difference between TwPabd and TwPes. The mean value of at least three signals was made to determine TwPdi on a given occasion.

Mean values of TwPdi, TwPabd, and TwPes in the total case series were calculated by using the mean value for patients that were measured on more than one occasion. The repeatability of measurement on the same occasion was evaluated by calculating the within-occasion coefficient of variation. Repeatability on different occasions was calculated by using data from patients who were measured on at least two different occasions and by determining the between-occasion coefficient of variation. The relation between TwPdi and duration of mechanical ventilation was evaluated by using regression analysis and applying the logarithmic model.

Informed consent was obtained in 25 patients (Figure 2). In eight of these patients, no stimulation was performed. Reasons were withdrawal of consent in one, inability to position gastric/esophageal catheters in three, technical problems with the valve in one, evolution of the medical condition by the time of the planned measurement, interfering with the planned measurement in three (moribundus, n = 1; unstable, n = 1; extubated, n = 1). In seven of 17 remaining patients, stimulation was performed on at least two occasions with a mean interval of 51.4 ± 35.1 (SD) h between measurements. One of these patients appeared to have unilateral diaphragm paralysis after surgery, confirmed by a phrenic nerve-conduction study, and another patient was measured on 4 consecutive days. In three other patients, measurements were made on a single occasion. In the remaining seven patients, no TwPdi values are available because of intolerance, technical problems, or active abdominal contraction during the experiment.

The reason for admission to ICU and the demographic data of the patients are shown in Table 1. The age ranged from 35 to 78 years. Measurements were made between 2 to 30 days after admission to ICU. All but one patient had sepsis, and except for one patient, all of these received vasopressors. Only one patient received renal-replacement therapy. Three patients were treated with aminoglycosides, and half of the patients received at least one bolus of neuromuscular blocking agents. All measurements were made at a PEEP level of between 5 and 7 cm H ₂O. Patients that were measured on more than one occasion were measured at the same PEEP level. Supramaximal stimulation was reached in 13 of 19 TwPdi measurements; in two of 19, supramaximality was not reached; and in the remaining four of 19, insufficient data are available to evaluate supramaximality.

Mean values of TwPdi, TwPabd, and TwPes in the total case series were calculated by using the mean value for the seven patients that were measured on more than one occasion. The TwPdi was 11.5 ± 3.9 cm H ₂O (mean ± SD); the TwPabd was 5.8 ± 2.3 cm H ₂O (mean ± SD); and the TwPes was 6.7 ± 2.8 cm H ₂O (mean ± SD). The within-occasion coefficient of variation was 7.25% and was calculated by using all accepted signals for all measurements (Table 2). No data are available for patient 3 in this table, as analysis revealed only one acceptable tracing for this patient. The mean between-occasion coefficient of variation was calculated by using the data of patients measured on at least two occasions and was 9.7% (Table 3).

Regression analysis, excluding the patient with unilateral diaphragm paralysis after surgery, suggested a logarithmic relation between the duration of mechanical ventilation and TwPdi (see Figure 3), with P = 0.038 and R = 0.69. To examine whether this effect might be due to other interfering variables, such as sedation or mode of mechanical ventilation, we also performed simple regression analysis for these variables. This suggested a logarithmic relation between TwPdi and the cumulative dose of piritramide up to the time of measurement (R = 0.79; P = 0.01); cumulative dose of propofol (R = 0.66; P = 0.05); duration of bilevel positive-pressure ventilation (BIPAP/ASB) (R = 0.71; P = 0.03); and pressure-support ventilation (PSV) (R = 0.66; P = 0.05), but no relation with duration of volume-controlled ventilation (IPPV) (see Figure 3).

This is the third case series reporting diaphragmatic force in critically ill patients by using BAMPS. Although the technique appears to have adequate repeatability to discriminate between moderately and severely reduced diaphragmatic force, some variability remains present and should be taken into account when interpreting a single value. This variability seems to be inherent to the method itself, as it also is present in healthy volunteers. In critically ill patients, changes of end-expiratory lung volume over time and between measurements are of particular concern. We attempted to minimize the potential impact by measuring patients at the same level of PEEP on different occasions and checked the flow curve to determine the end-expiration. We did not measure or correct for intrinsic PEEP. We cannot therefore exclude the possibility that changes may have occurred in end-expiratory lung volume between two measurements. As repeatability is very similar to that outside the ICU, the impact seems to have been limited.

We established a decrease of Pdi BAMPS with increasing duration of mechanical ventilation. It is therefore questionable to what extent repeatability of BAMPS Pdi between different occasions can actually be measured in critically ill patients. As we anticipated this possibility, we aimed at measuring patients in time intervals as close together as possible. We did not systematically find lower values on the second occasion than on the first occasion.

The technique is highly sophisticated, fairly invasive, and requires patients to be stable. For these reasons, the technique is currently limited in use to a small subgroup of critically ill patients. As currently no direct therapeutic measures are available to improve patients' respiratory muscle force, the technique only can be offered to patients and their relatives merely for research purposes. Obtaining informed consent for a relatively invasive procedure is therefore not obvious, especially in the early hours and days of admission to the intensive care unit, during which the prognosis of the patient is often unclear and the psychological burden for the relatives is high. Positioning the balloon catheters in patients who are not conscious is often very difficult, as these patients are not able to swallow, and catheters are very flexible and prone to curve in the nasopharynx, as well as in the esophagus. For these reasons, obtaining a large dataset of measurement is very difficult. The limited number of patients implies that we could not perform a stepwise regression analysis to determine the best predictor(s) of TwPdi. The relations that we have established are based on a small dataset, so the results may have been affected by other confounders and must be confirmed.